There are systems in which plural computers share and execute a processing. In such systems, changing allocations of processing during operation of the system may occur for the purpose of load balancing and the like.
Conventionally, in order to prevent a specific apparatus from being overloaded, there has been a technique for changing an apparatus that carries out the processing. More specifically, in a network in which plural managed apparatuses and plural management apparatuses are connected, each management apparatus counts the number of notifications from managed apparatuses that it manages per unit time for each managed apparatus. When the number of notifications becomes equal to or greater than a preset value, the management apparatus sends information regarding a management apparatus that will become a new notification destination to the managed apparatus, based on a management apparatus table that stores information about other management apparatuses. As a result, the load due to the processing that the management apparatus was originally supposed to carry out, or in other words, the processing caused by notifications from the managed apparatuses, is distributed.
Incidentally, when changing the allocation of the processing, in addition to the load due to the processing that was originally to be carried out, there is a load that occurs due to the processing to ensure that the processing after the allocation has been changed is carried out with no problems. For example, in the technique described above, a load occurs in the management apparatus due to the processing such as a processing carried out by the management apparatus to send information regarding the new destination management apparatus to the managed apparatuses.
On the other hand, there is a system in which information about the allocation destinations is not sent to apparatuses that will be affected by a change in allocations, when the allocation of the processing has been changed. In such a system, an apparatus that is affected by a change in allocations itself makes an inquiry for new allocation destination information to the system, and a load occurs in the system by handling such an inquiry. In the conventional techniques, reducing the loads that occur in the system when the allocations of the processing have been changed was not taken into consideration.
Namely, according to the conventional techniques, in a system that carries out and shares a processing by plural apparatuses, a load that occurs in the system when the allocations of the processing have been changed cannot be reduced.